// this procedure selects a leaf and a word and then splits the leaf and
// creates two new leaves
private void improve() {
Node bestLeaf = null; // getRandomLeaf();
String bestWord = null; // Main.dict.getRandomWord();
double bestIG = 0;
Node currentLeaf;
String currentWord;
ArrayList<String> words;
ArrayList<Message> msgs;
for (int i = 0; i < leaves.size(); i++) {
currentLeaf = leaves.get(i);
msgs = currentLeaf.getMessages();
for (int j = 0; j < msgs.size(); j++) {
words = msgs.get(j).getWords();
for (int k = 0; k < words.size(); k++) {
currentWord = words.get(k);
double currentIG = IG(currentWord, currentLeaf);
if (currentIG >= bestIG) {
bestIG = currentIG;
bestWord = currentWord;
bestLeaf = currentLeaf;
}
}
}
}
// System.out.println("Chose: "+ bestWord+" with IG: "+ bestIG);
split(bestLeaf, bestWord);
if (tenFirstWords.size() < 10) {
tenFirstWords.add(bestWord);
}
// improvement
leaves.remove(bestLeaf);
leaves.add(bestLeaf.getLeft());
leaves.add(bestLeaf.getRight());
}
private double N(Node L, int i) {
int sum = 0;
ArrayList<Message> msgs = L.getMessages();
for (int j = 0; j < msgs.size(); j++) {
if (msgs.get(j).getClassification() == i) sum++;
}
return (double) sum;
}
private void split(Node leaf, String word) {
leaf.setIsInnerNode(word);
Node newLeafLeft; // contains word
Node newLeafRight; // don't contain word
ArrayList<Message> messagesWithWord = new ArrayList<Message>();
ArrayList<Message> messagesWithOutWord = new ArrayList<Message>();
ArrayList<Message> messages = leaf.getMessages();
Message msg;
for (int i = 0; i < messages.size(); i++) {
msg = messages.get(i);
if (msg.contains(word)) {
messagesWithWord.add(msg);
} else {
messagesWithOutWord.add(msg);
}
}
newLeafLeft = new Node(messagesWithWord);
newLeafRight = new Node(messagesWithOutWord);
leaf.setLeft(newLeafLeft);
leaf.setRight(newLeafRight);
}
private double N(Node L) {
return L.getMessages().size();
}
